Stampedes have been common to populous India, with most occurring during religious gatherings. New studies in 'pedestrian dynamics' now say that to prevent a stampede, one needs to spot the building of pressure in a dense moving crowd.

Dirk Helbing of Swiss Federal Institute of Technology, Zurich, and Anders Johansson of the Centre for Advanced Spatial Analysis, University College, London, found that the decisive factor leading to a stampede was not average or local crowd density but pressure, which is speed variance multiplied by density.

Replying to an e-mail query, Johansson told the media: "For planned mass gatherings like pilgrimages, sports and political events, a set of measures should be taken in order to minimise the risk of a crowd crush or other types of disasters.

"Firstly, one has to make sure that the infrastructure (roads, corridors, open spaces, entrances and exits) is suitable for the mass gathering so that there is enough capacity and there are no bottlenecks or other compression points.

"Secondly, there should be a good crowd-management plan that is followed, which includes crowd monitoring, scheduling and control. Thirdly, one must have good contingency plans (e.g. evacuation) in case anything goes wrong."

"On top of this, it is advisable to make use of modern technology such as computer simulation of crowds, to test the suitability of a venue virtually in a computer, before the gathering, and also to use any mean of real-time crowd counting during the event, to be able to take action in case the event does not unfold according to plan," he said.

India has a long history of stampede tragedies.

On November 8, nearly 20 people were trampled to death in Haridwar, Uttarakhand, during a yagna. On January 14, over 102 people were killed in a stampede near the shrine of Sabarimala in Kerala. This was the second accident there since 1999. On March 4, 2010, 63 deaths occurred at Ram Janki Temple in Kunda, Uttar Pradesh.

On September 30, 2008, at the Chamunda Devi temple in Jodhpur, Rajasthan, 249 people were killed. On August 3 the same year, 163 were crushed to death along the path leading to the Naina Devi Temple in Himachal Pradesh. There have been stampedes at recruitment drives of the army and at various other types of gatherings.

Helbing and Johansson made their discovery while studying the stampede during the Haj in Mecca-Medina on January 12, 2006 - how propulsion force of people increased in areas of extreme densities. The next year, implementing their suggestions for corrections, the Saudi Arabian government organised a safe Haj.

The two scientists had mathematically analysed the video recordings of the stoning ritual, and seen how, unlike vehicle traffic, even at extreme densities people never stopped moving. This led to self-generated deathly waves or ‘turbulence’ and an inevitable stampede. Scientifically, this was a new understanding and applicable to other mass gatherings.

Physicists until then had identified only two forms of dense crowd movements - laminar or streamlined flow (smooth, without hindrances) and stop-and-go flow (like movement of vehicular traffic) using models such as ‘directional segregation’ and ‘cellular automata’.

A study of the Haj stampede showed that when people are so close to one another that it becomes unbearable and asphyxiating, the pressure makes them push out to gain space, leading the crowd to the state of ‘turbulence’. Then the crowd, carrying up to nine people per square metre, resembling a fluid mass, begins to sway like a wave and finally ‘ruptures’, leading to ‘crowd panic’.

Like tremors of an earthquake, the situation becomes uncontrollable, moves beyond the power of security forces, if any. The mass splits up into clusters, spreading with strong velocity in all possible directions. People are pushed around, some stumble and fall and, if they cannot get up fast, are trampled upon. Soon, the area of trampled people becomes larger and larger as they become obstacles to others. The crowd has unknowingly self-destructed itself.

To stop the crowd from reaching this stage, we need advance warning signals of ‘critical crowd conditions’. The signals can be evaluated online by an automated video analysis system and directions can be sent out to control the crowd.

This can be through crowd flow control, pressure relief strategies or separation of people into blocks to stop the spreading of shock waves.

Pedestrian dynamics as a field came into being in 1995 and physicists since then have contributed to the understanding of crowd disasters, each moving closer to the heart of the imminent tragedy but not quite reaching it.